- Balanced AI Insights
- Posts
- AI News of the Month — August 2025
AI News of the Month — August 2025
ChatGPT-5 and gpt-oss models, Anthropic data policy, AI browsers, ESG impacts, and more.

🚀 Big News: I’m launching my new masterclass — AI for CFOs: The Definitive Leadership Masterclass
Starting September 30th, I’ll be leading a 6-week live virtual program together with CAPRUS AI. This isn’t another “AI hype” course. It’s a CFO-built, CFO-led roadmap designed to cut through the noise and give finance leaders the tools, governance frameworks, and ROI models to lead with confidence.
Over six weeks, we’ll move from AI fundamentals to investment strategy, vendor management, and executive roadmaps — with frameworks you can implement immediately. Whether you’re scaling a startup, steering a mid-market firm, or governing a global enterprise, you’ll leave with one thing every CFO needs right now: a personalized, practical AI adoption plan. Cohort 1 begins September 30th — secure your seat and join us in shaping the next chapter of financial leadership.
AI News of the Month — August 2025
1. OpenAI’s August Updates — GPT-5 & Open-Source Models
OpenAI released GPT-5, one of the most anticipated launches of the year. While the model brought notable improvements in reasoning and multimodal workflows, many teams reported a short period of disruption as prompts, workflows, and integrations needed to be re-tuned.
More quietly but strategically, OpenAI also released the “gpt-oss” models, an open-source model — a move that could reshape enterprise adoption by allowing safer, self-hosted deployments.
Read more
So What:
GPT-5 highlights a core reality: finance leaders rely on AI models that keep changing under the hood. Even small shifts can ripple through reporting workflows and team processes. But the open-source release is the bigger story for CFOs: it opens a path to enterprise-hosted AI environments, with stronger governance, data security, and cost control.
What to Do:
Expect periodic workflow adjustments whenever models update — build change-management capacity into your AI adoption strategy.
Begin exploring self-hosted or vendor-hosted open-source models for sensitive use cases where governance and compliance matter most.
Ask IT and vendors whether they are testing the new OpenAI open-source stack — it may offer a safer, more cost-effective path than consumer subscriptions.
2. Anthropic’s Consumer Data Policy Shift
Anthropic now requires Claude Free/Pro/Max users to actively opt in or out of data being used for training by Sept 28, 2025. If you don’t choose, you’ll lose access. Enterprise and API accounts remain unaffected.
Read more
So What:
Many solo practitioners and fractional CFOs use Claude Pro for real client work. Unless you opt out, your prompts and business data will automatically feed Anthropic’s training pipeline.
What to Do:
If you’re on a Free/Pro/Max plan, log in and opt out of training — this protects client data and unlocks more safe use cases.
If you work in a team, consider moving to Claude Team or Enterprise for stronger governance.
Expect other vendors to follow a similar path — collecting user data is tempting, and the default will rarely favor your privacy. ChatGPT has always operated this way. Don’t ignore these notices; take every chance to protect your data.
3. AI-Driven Browsers — Claude for Chrome
Claude for Chrome now clicks, fills forms, and navigates websites. Perplexity is building similar features with its browser Comet.
Read more
So What:
The idea of an AI agent that clicks through portals and pulls reports is compelling. But current tests show double-digit rates of successful attacks (prompt-injection success rates of 23.6%). That makes these tools exciting for experimentation, but not yet safe for finance workflows. Leaders should track them now to be ready when security catches up.
What to Do:
Pilot only in safe sandboxes (e.g., downloading PDFs from public sources).
Keep AI browsers completely separate from financial systems, payments, or sensitive portals.
Update your AI usage policy to note that AI browsers are emerging tools — they deserve a place in your roadmap, but only for monitored experimentation.
Begin documenting potential safe workflows (like report retrieval or vendor portal navigation) for future adoption once security improves.
4. AI and the Environment — Transparency by the Byte
Google published a report disclosing the environmental impact of running AI queries. Here are the main stats:
Median text prompt energy use: 0.24 Wh (≈1 second of microwave).
Water use per prompt: 0.26 ml (≈5 drops).
Carbon footprint per prompt: 0.03 g CO₂ (Google’s clean energy–adjusted).
Efficiency: 33× reduction from May 2024 → May 2025.
So What:
ESG-minded boards are increasingly asking about AI’s footprint. I’m delivering an AI training for a nonprofit next week, and they specifically asked me to cover the environmental impact. As I prepared, I realized it’s not nearly as terrible as headlines suggest — and it’s improving quickly. The takeaway: AI’s environmental costs are real but manageable, and with efficiency gains, the trend is moving in the right direction. Transparency is finally catching up.
What to Do:
Add AI’s energy and water usage to ESG reporting frameworks.
Ask your cloud vendors for sustainability disclosures.
Frame AI adoption as net-positive: efficiency gains vs. environmental costs.
5. Platform Roundup — GPT-5, Microsoft Copilots & QuickBooks Agents
Microsoft advanced Dynamics 365 Finance Copilot, making natural-language ERP queries more powerful.
QuickBooks expanded its AI agents for reconciliations, auto-categorization, and monthly trend monitoring.
Oracle rebranded its annual flagship conference from CloudWorld to AI World — a symbolic but powerful signal that AI is now the centerpiece of its enterprise strategy.
So What:
Core finance platforms are embedding AI directly into their workflows. This isn’t experimental anymore — your ERP, accounting, and reporting systems are steadily becoming AI-native. The challenge is keeping adoption structured so governance doesn’t lag behind.
What to Do:
Ask ERP/accounting vendors which new AI features are on the roadmap for Q4.
Pilot one low-risk feature (variance explanations, auto-reconciliation) and measure results.
Build a quarterly review process for enabling or deferring new AI features in core systems.
Closing Thoughts
That’s the lineup of AI news for August. Summer is officially behind us, and fall is shaping up to bring a flood of new updates, releases, and events.
Stay tuned — I’ll be sharing some exciting announcements with you soon.
And thank you to everyone who voted in last week’s poll. You asked me to put more emphasis on AI use cases, so this week’s vote is all about you: which finance use cases are you most curious to explore next? Cast your vote, and I’ll build the next edition around the winner.
Which real-world AI adoption example are you most interested in seeing explored next? |
We Want Your Feedback!
This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!
Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!
Did you find this newsletter helpful? Forward it to a colleague who might benefit!
Until next Tuesday, keep balancing!
Anna Tiomina
AI-Powered CFO
Reply